Impact of CPU-bound Processes on IP Forwarding of Linux and Windows XP

These days, commodity-off-the-shelf (COTS) hardware and software are used to build high-end and powerful workstations and servers to be deployed in today's local area networks of private homes and small- to medium-sized business. Typically, these servers are multipurpose and shared - running networking functionalities involving IP packet forwarding in addition to other CPU intensive applications. In this paper we study and investigate the impact of running CPU-bound applications on the performance of IP packet forwarding. We measure and compare the impact and performance for the two operating systems of choice for home and small-business users, namely Linux and Windows XP. The performance is studied in terms of key performance metrics which include throughput, packet loss, round-trip delay, and CPU availability. For our measurements, we consider today's typical home network hosts of modern processors and Gigabit network cards. We also consider different configuration setups and utilize open-source tools to generate relatively high traffic rates. Our empirical results show that Linux exhibits superior performance over Windows XP in terms of IP forwarding performance. Results also show that, unlike Windows XP, the IP forwarding performance of Linux is not significantly impacted by running CPU-bound applications.

[1]  H. Saleem,et al.  Performance comparison of packet transmission over IPv6 network on different platforms , 2006 .

[2]  Stefan Lundberg,et al.  Long-term viability of smart home systems - Business modelling and conceptual requirements on technology , 2005 .

[3]  G. Prytz,et al.  Real-time performance measurements using UDP on Windows and Linux , 2005, 2005 IEEE Conference on Emerging Technologies and Factory Automation.

[4]  Timothy S. Newman,et al.  Performance Comparison , 2021, Satellite Formation Flying.

[5]  EDDIE KOHLER,et al.  The click modular router , 2000, TOCS.

[6]  Sebastian Zander,et al.  KUTE A high performance Kernel-based UDP traffic engine , 2005 .

[7]  Daniel Pierre Bovet,et al.  Understanding the Linux Kernel , 2000 .

[8]  Scott O. Bradner,et al.  Benchmarking Methodology for Network Interconnect Devices , 1999, RFC.

[9]  Dror G. Feitelson,et al.  Comparing Windows NT, Linux, and QNX as the basis for cluster systems , 2001, Concurr. Comput. Pract. Exp..

[10]  Andry Rakotonirainy,et al.  A Survey of Research on Context-Aware Homes , 2003, ACSW.

[11]  Kenji Takeda,et al.  Comparative performance of a commodity Alpha cluster running Linux and Windows NT , 1999, ICWC 99. IEEE Computer Society International Workshop on Cluster Computing.

[12]  Eddie Kohler,et al.  The Click modular router , 1999, SOSP.

[13]  Sherali Zeadally,et al.  Comparison of end-system IPv6 protocol stacks , 2004 .

[14]  Edv Informatik.  Apache HTTP Server , 2010 .

[15]  Jamal Hadi Salim,et al.  Beyond Softnet , 2001, Annual Linux Showcase & Conference.

[16]  Sherali Zeadally,et al.  Network application programming interfaces (APIs) performance on commodity operating systems , 2004, Inf. Softw. Technol..

[17]  Wen-Shyang Hwang,et al.  A QoS-aware residential gateway with bandwidth management , 2005, IEEE Transactions on Consumer Electronics.

[18]  Lorenzo Muttoni,et al.  Performance evaluation of a Windows NT based PC cluster for high performance computing , 2004, J. Syst. Archit..

[19]  Khaled Salah,et al.  Evaluating system performance in Gigabit networks , 2003, 28th Annual IEEE International Conference on Local Computer Networks, 2003. LCN '03. Proceedings..

[20]  Khaled Salah,et al.  Comparative packet-forwarding measurement of three popular operating systems , 2009, J. Netw. Comput. Appl..